Objective: To assess the accuracy of plasma N-terminal-pro-B-type natriuretic peptide concentrations (NT-proBNP) as a diagnostic tool to recognize acute respiratory failure (ARF) of cardiac origin.
Methods: Prospective observational study in 100 medical intensive care unit (ICU) patients. NT-proBNP was measured at ICU admission, and diagnosis of cardiac dysfunction was performed using echocardiography.
Results: Sixteen patients had cardiac ARF, 58 patients had noncardiac ARF, and 26 patients were non-ARF controls. Median (IQR) NT-proBNP was 1,951 (617-9,320) pg/ml and was significantly influenced by the level of renal dysfunction. Patients with noncardiac ARF had higher NT-proBNP [1,912 (704-1,922) pg/ml] than non-ARF patients [1,022 (383-2,613) pg/ml], but lower concentrations than cardiac ARF patients [4,536 (1,568-35,171) pg/ml]. The area under the curve (AUC) was 0.663+/-0.078 (95% confidence interval 0.510-0.815) and was not significantly influenced by the level of renal dysfunction. In addition, using a stepwise logistic regression model, NT-proBNP failed to predict independently the presence of cardiac dysfunction. However, with specificity and negative predictive value of 100%, a NT-proBNP cutoff value of 500 pg/ml seemed useful to rule out cardiac dysfunction. Indeed, none of the 16 patients with cardiac ARF had a NT-proBNP value below 500 pg/ml, whereas it was the case in 8 (30.8%) non-ARF controls and in 12 (20.7%) noncardiac ARF patients.
Conclusions: In cancer patients with ARF, plasma NT-proBNP concentration is not a relevant tool to recognize cardiac dysfunction, but is specific enough to rule out the diagnosis in patients with plasma NT-proBNP concentrations below 500 pg/ml.